Investigation of near-infrared light propagation with Monte Carlo and phantom simulations

Zhengyu Jiang

doi:10.17918/00009270

Near-infrared spectroscopy (NIRS) has been widely used to noninvasively measure changes in the concentrations of oxy- and deoxyhemoglobin in tissue. The exact nature of light propagation in brain tissue, however, remains elusive. Effective and realistic modeling of light propagation in human head holds enormous potential for researchers to better understand capabilities of NIRS technology in neuroimaging applications. Currently there exists several mathematical (deterministic or stochastic) models of light transport in tissue. Within those models, Monte Carlo simulations" are widely used due to its versatility in handling complex geometries, inhomogeneities, as well time-dependency through simple implementation. In this thesis, light propagation is first studied by Monte Carlo simulations, and then validated by experimental adult human head mimicking laboratory phantom model measurements. An existing Monte Carlo simulation program is taken as the basis in this study and several updates are incorporated to make the model represent the light propagation in tissue more realistically and to provide several different features as the outcome of the program. The simulations in this study, consisting of either a homogeneous four-layered head model or a heterogeneous five-layered head model, are set up to systematically demonstrate the changes of optical pathlength, differential pathlengh factor (DPF) and penetration depth, three-dimensional historical paths and corresponding spatial sensitivity profile (SSP) as a function of different source-detector separations and at different wavelengths of light. Furthermore, effects of the thicknesses of various superficial head layers such as scalp, skull and cerebrospinal fluid (CSF) on mean and partial pathlength and penetration depth of the light through the overall head and within the brain are investigated. The Monte Carlo simulations reveal that 1) DPF values obtained from simulation are within acceptable ranges as compared to prior simulation studies, albeit smaller than those from real measurements, 2) penetration depth is not wavelength dependent and is positively correlated to mean optical pathlength, 3) SSP can not be directly found from 3D historical paths even though there exist a relationship between them, and 4) the thickness of the scalp at the most superficial layer of the head has the strongest effect on mean optical pathlength and penetration depth whereas the thickness of the skull has the strongest effect on partial pathlength and penetration depth in the brain and the thickness of CSF which can be considered as a clear layer has no effect on the mean optical pathlength. In addition, the simulations from the four-layered head model are consistent with those obtained from experimental phantom model results-in which mean optical pathlength is correlated with penetration depth. However, heterogeneous five layered head model simulation results further suggest that mean optical pathlength and penetration depth of light through the overall head does not fully represent the partial pathlength and penetration depth of light within the brain itself since variations in the thicknesses of various superficial layers can result in differences in near infrared light propagation within those layers and hence in the brain as well

Investigation of near-infrared light propagation with Monte Carlo and phantom simulations

Files and links (1)

Abstract

Metrics

Details

Investigation of near-infrared light propagation with Monte Carlo and phantom simulations

Files and links (1)

Abstract

Metrics

Details

Drexel University Social media